How to Stop Being the Answer to Every Question

Why this matters

The interruption is never the cost. A tech calls to ask what you charge for a standard diagnostic, you answer in ninety seconds, and the real damage is the forty minutes it takes to get back into the estimate you were halfway through. Do that seven times a day and the estimate goes out Saturday morning instead of Thursday afternoon, which is why you are working weekends while telling yourself the week was not that bad.

There is a second cost that only shows up when you are unreachable. If every answer lives in your head, the shop stalls the day you are under a house with no signal. That is not a delegation problem or a staffing problem. It is a retrieval problem, and it is fixable without hiring anybody.

Separate a question from an approval before you start

These get lumped together and they have different fixes. An approval is a decision someone already knows the answer to but lacks the authority to make: they know the part needs replacing, they need you to say yes to the spend. A question is one where the person genuinely does not have the information or the judgment: what do we charge, how long is our labor warranty, does this job need a permit.

Approvals are fixed with authority limits and written thresholds, and a sibling article already owns that ground. This article is the other half: the questions. Sorting the two at the top matters because handing out more authority does nothing for a tech who simply cannot find the price list.

Step 1: Log every question for two full weeks

One week is a slice of a season, not a pattern. Ten working days gives you a median you can trust and enough repeats to see which questions are actually recurring rather than coincidental.

Five fields, kept on your phone or a notepad in the truck:

That last column is the one people skip and it is the one that changes behavior. Skip it and you conclude the interruptions cost a couple of minutes each, which is why owners who log without it decide the problem is not worth fixing and stop.

Log everything, including the ones you enjoy answering. Especially those. A question you like answering is the hardest kind to give away and it is usually the one arriving most often.

Step 2: Classify by how long the answer stays true

Every logged question falls into one of three classes, and the class determines the fix. Do not sort by who asked or by topic.

• Lookup. The answer exists, it is the same today as last month, and the person just cannot find it. What do we charge for a standard visit. What is our warranty term. Which supplier do we use for that fitting. Fix: publish it where they already look.
• Rule. The answer is the same every time but nobody has ever written the rule down, so it lives as a habit in your head. When do we waive a trip charge. When do we call the customer before proceeding. Fix: write the rule with its threshold and its Boolean.
• Judgment. The answer genuinely moves with the job and needs experience to reach. Fix: coach the reasoning and set a check-back, not an answer.

The classification test: if you asked the same question on a different job next month and the answer would change, it is Judgment. If it would not change, it is Lookup or Rule, and the only difference between those two is whether an answer already exists somewhere in writing.

Step 3: Publish the lookups before you touch anything else

Lookups are usually the largest class and the cheapest to kill. The trigger to publish, stated with its unit: any distinct question that appears three or more times across the ten day log, counted across all askers rather than per person, becomes a published answer. Three is the point where you are looking at a pattern and not a bad week for one person.

The gate that goes with it is an AND, not an OR: publish it when it hits three or more occurrences and the answer does not move with the job. A question that hits three occurrences with a moving answer is Judgment wearing a Lookup costume, and publishing a fixed answer to it will produce wrong work.

Publish where they already look, not where you think they should look. If your crew lives in text messages, a shared document nobody opens is not published. One page pinned in the app they use daily beats a well-organized folder they do not.

Step 4: Write the rules with a threshold, a Boolean and a step

A rule you state as a principle comes straight back to you as a question, because the person cannot tell whether their case qualifies. "Use your judgment on trip charges" is not a rule. "Waive the trip charge when the drive is under fifteen minutes from the previous stop and the customer has been with us over a year" is a rule, because it names both gates and joins them.

Three things every written rule needs:

1. The unit of analysis. Per job, per customer, per week, per truck. "Over two hours" means nothing until you say two hours of what, measured across what.
2. The Boolean when there are two gates. AND or OR, stated. Most rules that come back as questions came back because the writer left this out and the reader guessed.
3. The step, if the rule changes a number. If the rule says raise the quoted hours when a job type runs over, say by how much and off what baseline.

Then run one real past job through your own written rule before you hand it over. If your own recent decision does not survive contact with your own rule, the rule is wrong, not the decision.

Step 5: For judgment calls, coach the reasoning and set a check-back

Judgment questions are the ones you should still be getting, so do not try to eliminate them. Change what you hand back. Instead of the answer, hand back the two or three things you would look at, then set a check-back: they decide, they tell you what they decided and why, you correct only if it was wrong.

The check-back cadence that works without turning into supervision: end of day for the first four weeks after handing off a class of decision, then twice a week for four weeks, then only on exceptions. The step down happens on a date, not on a feeling, or it never happens.

The failure here is answering the judgment question anyway because you are in a hurry. That is the correct call once. Done three times, the person stops forming their own read and starts waiting for yours, and you have taught them to interrupt you.

A worked two-week log

An owner running four techs logged for ten working days and caught 74 questions. The interruption itself averaged 4 minutes. The recovery column averaged 9 minutes. So the true cost per question was about 13 minutes, and 74 times 13 minutes is 962 minutes, roughly 16 hours across the two weeks, about 8 hours a week.

Classified: 41 Lookup, 22 Rule, 11 Judgment. That is 41 of 74, about 55 percent of all logged questions, sitting in the cheapest class.

Inside the 41 lookups, only 9 distinct questions appeared three or more times, and those 9 accounted for 29 of the 41 lookup occurrences. Pricing on the three most common job types, the labor warranty term, the two suppliers, and three recurring parts questions. One page, published in the app the crew already opened every morning.

The 22 rule questions collapsed into 5 written rules, each with a threshold and a Boolean. The 11 judgment questions got a check-back instead of an answer.

Re-logged for ten working days at week eight: 38 questions. Lookup fell from 41 to 9. Rule fell from 22 to 12. Judgment rose from 11 to 17. At 13 minutes each, 38 questions is about 494 minutes, roughly 8 hours across the two weeks, about 4 hours a week, down from about 8 hours a week.

The judgment increase is the part to read correctly. It did not go up because the crew got worse. It went up because they stopped burying hard calls under easy ones and because two of them had started taking calls they would previously have punted. Reading that number as a failure and clamping down is the way to undo the whole exercise.

The retest that tells you a fix did not hold

The single diagnostic is repetition after publication. If a published lookup appears three or more times again in the next ten day log, the answer is not findable, and the fix is to move it, not to rewrite it. Owners reflexively rewrite: they add detail, they reorganize, they make the page longer. The page was never the problem. Location was.

Rewrite only when the question coming back is worded differently from the one you published. That is a comprehension miss, and it is fixed by matching the crew's wording, which is why the log captured their words rather than your summary.

The other thing worth watching: whether the same person accounts for a rising share of what is left. In this log, at week eight one tech accounted for 14 of the 38 questions, well over a third of the total. That is not a documentation problem, it is a training gap in one person, and it gets handled as coaching rather than by another published page.

References

• See related: Universal - The Owner Bottleneck: When Everything Needs My Approval (approvals and authority limits, the other half of this split)
• See related: Delegating the Decision, Not Just the Task
• See related: Universal - The Standing Question Log SOP (the running instrument this method feeds)
• See related: Batching vs Switching: The Cost of Interruptions
• U.S. Small Business Administration, guidance on documenting standard operating procedures for small businesses